From: Charles Oliver Nutter <headius@...>
Date: 2009-10-26T12:27:52+09:00
Subject: [ruby-core:26309] Re: IO.foreach (and friends) effect on $< and 	$.

On Sun, Oct 25, 2009 at 9:22 PM, Marc-Andre Lafortune
<ruby-core-mailing-list@marc-andre.ca> wrote:
> I believe there is a bug in trunk with $. but I am not certain how to
> resolve it.
>
> Is there an official documentation for $. (and all other global
> variables) in rdoc?
>
> The Ruby Prog Language (Flanagan & Matz) states:
>
> The number of the last line read from the current input file.
> Equivalent to ARGF.lineno.
>
> I find "current input file" to be very imprecise. Is that the file
> currently being read from ARGF? or the last IO read from?

It is definitely imprecise, especially now considering that 1.8 and
1.9 do things differently

> Currently, it's a bit of both. It's usually the last read io:
>
> $ rubydev -e "ARGF.gets; f=File.new '/etc/passwd'; 4.times{f.gets}; p
> $., ARGF.lineno" /etc/hosts
> 4
> 1

Results on my system; "ruby" is OS X 10.5 latest 1.8.6 version:

~/projects/jruby ��� ruby -e 'ARGF.gets; f=File.new "/etc/passwd";
4.times{f.gets}; p $., ARGF.lineno' /etc/hosts
4
4

~/projects/jruby ��� ruby1.8.7 -e 'ARGF.gets; f=File.new "/etc/passwd";
4.times{f.gets}; p $., ARGF.lineno' /etc/hosts
4
4

~/projects/jruby ��� ruby1.9 -e 'ARGF.gets; f=File.new "/etc/passwd";
4.times{f.gets}; p $., ARGF.lineno' /etc/hosts
4
1

> Clearly, it is not equivalent to ARGF.lineno in this case.
>
> On the other hand, it's affected by ARGF.lineno= (contrary to any
> other IO#lineno=), and vice-versa:
> $ rubydev -e "ARGF.lineno = 42; p $.; $. = 56; p ARGF.lineno"
> 42
> 56

Yes, this was confusing to me as well. I don't understand why these
are intertwined.

- Charlie